基于POMDP的认知无线网络次用户多时隙信道选择算法

POMDP-BASED CHANNEL SELECTION ALGORITHM FOR MULTI-SLOT SU SYSTEM IN COGNITIVE RADIO NETWORKS

为了最大化认知无线网络次用户系统的吞吐量,同时减少信道选择过程中的计算量,在多条主用户信道的次用户多时隙系统中,用半马尔科夫链POMDP(Partially Observable Markov Decision Process)理论刻画次用户信道选择过程,并提出对应的认知无线网络信道选择算法。在每一个时隙开始时,次用户选择部分信道进行感知,之后根据感知结果选择不同的传输功率接入信道,对于没有被感知的信道,次用户可以直接接入。对POMDP问题求解给出最优策略、次优策略。通过在不同仿真环境下对最优策略、次优策略以及任意策略进行数值分析比较,论证了该设计算法的有效性。

In order to maximise the throughput of secondary users （SU） system in cognitive radio networks and to reduce the calculation amount in channel selection process at the same time, in the paper we utilise partially observable Markov decision process （POMDP） theory to depict SUs’ channel selection process in SU multi-slot system in multiple primary users channels, and present the corresponding channel selection algorithm of cognitive radio networks. At the beginning of each time slot, SUs select part of the channels to perceive, and select different transmission power to access the channel afterwards based on perception results, for those channels haven’t been perceived, SUs can access directly. In the paper we also present the optimal policy and suboptimal policy for solutions of POMDP problem. Through numerical analysis and comparison on optimal policy, suboptimal policy and random policy under different simulation environments, we demonstrate the effectiveness of channel selection algorithm designed in the paper.